CN106776856A - A kind of vehicle image search method of Fusion of Color feature and words tree - Google Patents
A kind of vehicle image search method of Fusion of Color feature and words tree Download PDFInfo
- Publication number
- CN106776856A CN106776856A CN201611069889.4A CN201611069889A CN106776856A CN 106776856 A CN106776856 A CN 106776856A CN 201611069889 A CN201611069889 A CN 201611069889A CN 106776856 A CN106776856 A CN 106776856A
- Authority
- CN
- China
- Prior art keywords
- image
- color
- feature
- vision word
- word
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000004927 fusion Effects 0.000 title claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 5
- 230000001174 ascending effect Effects 0.000 claims abstract description 4
- 101150060512 SPATA6 gene Proteins 0.000 claims description 27
- 230000011218 segmentation Effects 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 6
- 238000013139 quantization Methods 0.000 claims description 6
- 238000009333 weeding Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims 1
- 230000008447 perception Effects 0.000 abstract description 4
- 238000011002 quantification Methods 0.000 abstract description 2
- 238000004040 coloring Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000013517 stratification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/783—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of Fusion of Color feature and the vehicle image search method of words tree,The vehicle sections of training image are extracted first,Then local feature is extracted to vehicle sections,It is vision word words tree using k means methods are layered by characteristic quantification,Then test image is extracted into body portion local feature,This feature is assigned to corresponding vision word again,And using the principle of perception Hash,Vision word is each mapped to Hash codes,Obtain the Hash codes sequence of image,Then the color characteristic of vehicle body image is extracted,Quantify and unify as a feature degree,According to the ascending sequence of color distortion,Color characteristic weight is set up according to ranking results,Calculate Hamming distance between image Hash codes sequence,Using color characteristic weight as Hamming distance weight coefficient,Calculate the final similarity of image,The method effectively save vehicle image retrieval time,Improve the retrieval rate of vehicle image,Application request can be met.
Description
Technical field
The present invention relates to a kind of Fusion of Color feature and the vehicle image search method of words tree, belong to image procossing and divide
Analyse the application in intelligent transportation system.
Background technology
Vehicle image retrieval is the important component of intelligent transportation system, is that public security system cracks the stolen grade correlation of vehicle
The important means of case.So the magnanimity bayonet socket high-definition image in urban transportation, the accuracy and height of vehicle image retrieval
Effect property is extremely crucial to solving a case in time.Image Retrieval Mechanism based on bag of words is the main flow of field of image search in recent years
A large amount of training image features (typically SIFT feature) are clustered and are mapped as vision word by method, the method by k-means
Set, constitutes vision word dictionary.Then the feature of test image is matched one by one and is quantified in vision word dictionary, obtained
The vision word histogram of image.It is this due to this process time consume of a large amount of characteristic quantifications to vision word is excessive,
Nist é r et al. are proposed using the method generation vision word words tree for being layered k-means clusters, efficiently solve search non-
The too slow problem of the quantizing process that stratification word brings, (term frequency-inverse document is frequently by TF-IDF models for the vision word of generation
Rate) weighting.Because the SIFT feature of visual vocabulary tree-model use is only with half-tone information, and have ignored the global face of image
Color characteristic, therefore, the colouring information of vehicle image is dissolved into vision word Matching Model by the present invention as image overall weights
In, so that the retrieval precision of vehicle image is improved, while present invention introduces hash algorithm principle is perceived, can effectively save big rule
Mould vehicle image retrieval time.
The content of the invention
The technical problems to be solved by the invention are to provide a kind of Fusion of Color feature and are retrieved with the vehicle image of words tree
Method, retrieval time of the invention shortens, and improves retrieval precision to a certain extent, disclosure satisfy that in real use to accuracy
With the requirement of real-time.
In order to solve the above technical problems, the technical solution adopted in the present invention is：A kind of Fusion of Color feature and words tree
Vehicle image search method, it is characterised in that including following part：
The establishment of S01, words tree：Body portion is first extracted to each image in training image storehouse, for weeding out the back of the body
The vehicle image of scape part extracts SIFT feature, has thus obtained a characteristic set F={ f (i) }.Then to characteristic set
F carries out layering k-means clusters.When initial, on the ground floor of words tree carrying out first time k-means to characteristic set F gathers
Class, calculates the center vector C of each clusteri.Similar, k is polymerized to k-means again to new each cluster for producing
Gathering, is repeated continuously aforesaid operations until the depth set reaches L values set in advance, and each gathering is defined as a vision list
Word, obtains vision word words tree；
S02, represent image with weight vector：The body portion of image in test image storehouse is extracted, for weeding out the back of the body
The vehicle image of scape part extracts SIFT feature, and each SIFT feature is distributed into closest vision list in words tree respectively
Word, the number of times for then occurring according to vision word in each image and vision word calculate image with the degree of correlation of image
Vision word weights；
The calculating of S03, color characteristic weight：Image body portion hsv color spatial model is extracted, using homogenous segmentations
Method quantifies to feature, then will in characteristic vector unification to a feature degree, calculate query image and image to be retrieved it
Between feature degree Euclidean distance sequence and label, color characteristic weight is set up according to color notation；
S04, the selection for perceiving Hash system number：To improve the recall precision of large-scale image, the present invention is breathed out using perception
Uncommon algorithm principle, vision word is quantified respectively to different Hash codes, is considering quantization time of the invention with retrieval
After accuracy, system number of the invention is chosen for 4；
S05, image similarity are calculated：The perceived distance of image Hash sequence is calculated by Hamming distance, then color is special
Levy weight as the weight coefficient of Hamming distance obtain similarity sim (q, d) of final query image and image to be retrieved=
CqdHqd, will similar image output the most.
Further illustrated as of the invention, the step S02 is specially：
(1) to extracting body portion first per piece image in test library, SIFT feature is extracted to body portion, thus
A characteristic set F={ f (i) } and corresponding image ID set imgID={ id (i) } are obtained；
(2) SIFT feature is distributed to by closest vision word in words tree, so every width according to closest principle
The SIFT feature of image has been converted to vision word；
(3)wi,jThen represent image djMiddle vision word FiWeights, i.e. vision word FiWith image djDegree of correlation.Power
Value is defined according to the principle of TF-IDF, mi,jRepresent vision word FiIn image djThe number of times of middle appearance, it can be used to weigh
Amount vision word is used for describing the fine or not degree of image.N represents the sum of image in image library, niRepresent and include vision word Fi
Picture number.Inverse document frequency is defined as idfi=lg (N/ni), it represents and is meant that vision word for distinguishing similar diagram
The effect size of picture and dissimilar image.So, wi,jCan be expressed as
(4) image djD can be expressed as with the weight vector of vision wordj=[w1,j,w2,j,L,wt,j], test image
All images in storehouse can be expressed in matrix as
As a further improvement on the present invention, the step S03 is specially：
(1) HSV space color feature vector is directly extracted to image, in order to reduce computational complexity, using uniform point
The method of section quantifies to color characteristic, in order to reduce computational complexity, using the method for homogenous segmentations to color characteristic
Quantified, according to formula L=HQSQV+SQV+ V by characteristic vector unification to a feature degree, wherein QS、QVRepresent respectively
The quantization series of saturation degree and luminance component；
(2) the feature degree Euclidean distance between query image and image to be retrieved is calculated, it is ascending according to Euclidean distance
Be ranked up and label, color it is most like marked as 1, second is similar marked as 2, by that analogy；
(3) label according to color sets up color characteristic weight, the color characteristic weight calculation of query image q and image d
Formula isWherein RqdRepresent when query image is q, the label of image d.
Further illustrated as of the invention, the step S04 is specially：
(1) in view of system number of the invention is 4, the sum of same vision word weights in each image is calculated, averagely will
It is divided into four sections, and each segment table shows a Hash codes, obtains four Hash codes intervals；
(2) the Hash codes interval of the corresponding vision word of each vision word is compared, respectively regards each
Feel that word is mapped to Hash codes, obtain the Hash codes sequence of image.
The beneficial effects of the invention are as follows the color characteristic of image to be added to the weights of vision word as global weights
In matching somebody with somebody, global characteristics and local feature are considered, and introduce perception hash algorithm principle, in effectively save retrieval time
On the basis of, the accuracy rate of vehicle image retrieval is also improve to a certain degree.
Brief description of the drawings
Fig. 1 is the vehicle image search method flow chart of a kind of Fusion of Color feature provided by the present invention and words tree.
Specific embodiment
The present invention is described in detail for shown each implementation method below in conjunction with the accompanying drawings, but it should explanation, these
Implementation method not limitation of the present invention, those of ordinary skill in the art according to these implementation method institutes works energy, method,
Or equivalent transformation or replacement in structure, belong within protection scope of the present invention.
The SIFT feature used for the vehicle image search method for being currently based on bag of words and words tree is based only on
Half-tone information, have ignored the colouring information of image, and the present invention proposes the vehicle image of a kind of Fusion of Color feature and words tree
Search method, is specifically described below：
In the present embodiment, a kind of vehicle retrieval method of Fusion of Color feature and words tree, it includes following part：
The establishment of S01, words tree：Body portion is first extracted to each image in training image storehouse, for weeding out the back of the body
The vehicle image of scape part extracts SIFT feature, has thus obtained a characteristic set F={ f (i) }.Then to characteristic set
F carries out layering k-means clusters.When initial, on the ground floor of words tree carrying out first time k-means to characteristic set F gathers
Class, calculates the center vector C of each clusteri.Similar, k is polymerized to k-means again to new each cluster for producing
Gathering, is repeated continuously aforesaid operations until the depth set reaches L values set in advance, and each gathering is defined as a vision list
Word, obtains vision word words tree；
S02, represent image with weight vector：The body portion of image in test image storehouse is extracted, for weeding out the back of the body
The vehicle image of scape part extracts SIFT feature, and each SIFT feature is distributed into closest vision list in words tree respectively
Word, the number of times for then occurring according to vision word in each image and vision word calculate image with the degree of correlation of image
Vision word weights；
The step S02 is specially：
(1) to extracting body portion first per piece image in test library, SIFT feature is extracted to body portion, thus
A characteristic set F={ f (i) } and corresponding image ID set imgID={ id (i) } are obtained；
(2) SIFT feature is distributed to by closest vision word in words tree, so every width according to closest principle
The SIFT feature of image has been converted to vision word；
(3)wi,jThen represent image djMiddle vision word FiWeights, i.e. vision word FiWith image djDegree of correlation.Power
Value is defined according to the principle of TF-IDF, mi,jRepresent vision word FiIn image djThe number of times of middle appearance, it can be used to weigh
Amount vision word is used for describing the fine or not degree of image.N represents the sum of image in image library, niRepresent and include vision word Fi
Picture number.Inverse document frequency is defined as idfi=lg (N/ni), it represents and is meant that vision word for distinguishing similar diagram
The effect size of picture and dissimilar image.So, wi,jCan be expressed as
(4) image djD can be expressed as with the weight vector of vision wordj=[w1,j,w2,j,L,wt,j], test image
All images in storehouse can be expressed in matrix as
The calculating of S03, color characteristic weight：Image body portion hsv color spatial model is extracted, using homogenous segmentations
Method quantifies to feature, then will in characteristic vector unification to a feature degree, calculate query image and image to be retrieved it
Between feature degree Euclidean distance sequence and label, color characteristic weight is set up according to color notation；
The step S03 is specially：
(1) HSV space color feature vector is directly extracted to image, in order to reduce computational complexity, using uniform point
The method of section quantifies to color characteristic, in order to reduce computational complexity, using the method for homogenous segmentations to color characteristic
Quantified, according to formula L=HQSQV+SQV+ V by characteristic vector unification to a feature degree, wherein QS、QVRepresent respectively
The quantization series of saturation degree and luminance component；
(2) the feature degree Euclidean distance between query image and image to be retrieved is calculated, it is ascending according to Euclidean distance
Be ranked up and label, color it is most like marked as 1, second is similar marked as 2, by that analogy；
(3) label according to color sets up color characteristic weight, the color characteristic weight calculation of query image q and image d
Formula isWherein RqdRepresent when query image is q, the label of image d.
S04, the selection for perceiving Hash system number：To improve the recall precision of large-scale image, the present invention is breathed out using perception
Uncommon algorithm principle, vision word is quantified respectively to different Hash codes, is considering quantization time of the invention with retrieval
After accuracy, system number of the invention is chosen for 4；
The step S04 is specially：
(1) in view of system number of the invention is 4, the sum of same vision word weights in each image is calculated, averagely will
It is divided into four sections, and each segment table shows a Hash codes, obtains four Hash codes intervals；
(2) the Hash codes interval of the corresponding vision word of each vision word is compared, respectively regards each
Feel that word is mapped to Hash codes, obtain the Hash codes sequence of image.
S05, image similarity are calculated：The perceived distance of image Hash sequence is calculated by Hamming distance, then color is special
Levy weight as the weight coefficient of Hamming distance obtain similarity sim (q, d) of final query image and image to be retrieved=
CqdHqd, will similar image output the most.
Below the present invention is disclosed with preferred embodiment, so it is not intended to limiting the invention, all use equivalents
Or the technical scheme that equivalent transformation mode is obtained, it is within the scope of the present invention.
It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie
In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be in other specific forms realized.Therefore, no matter
From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the present invention is by appended power
Profit requires to be limited rather than described above, it is intended that all in the implication and scope of the equivalency of claim by falling
Change is included in the present invention.Any reference in claim should not be considered as the claim involved by limitation.
In addition, although the present specification is described in terms of embodiments, but not each implementation method is only only comprising one
Vertical technical scheme, this narrating mode of specification is only that for clarity, those skilled in the art should be by specification
As an entirety, technical scheme in each embodiment can also through appropriately combined, formed it will be appreciated by those skilled in the art that
Other embodiment.
Claims (4)
1. the vehicle image search method of a kind of Fusion of Color feature and words tree, it is characterised in that comprise the following steps：
The establishment of S01, words tree：Body portion is first extracted to each image in training image storehouse, for weeding out background portion
The vehicle image for dividing extracts SIFT feature, has thus obtained a characteristic set F={ f (i) }, and then characteristic set F is entered
Row layering k-means clusters, when initial, first time k-means cluster are carried out on the ground floor of words tree to characteristic set F,
Calculate the center vector C of each clusteri, it is similar, k cluster is polymerized to k-means again to new each cluster for producing
Collection, is repeated continuously aforesaid operations until the depth set reaches L values set in advance, and each gathering is defined as a vision list
Word, obtains vision word words tree；
S02, represent image with weight vector：The body portion of image in test image storehouse is extracted, for weeding out background portion
Point vehicle image extract SIFT feature, each SIFT feature is distributed into closest vision word in words tree respectively,
Then the number of times and vision word for occurring according to vision word in each image calculate image with the degree of correlation of image
Vision word weights；
The calculating of S03, color characteristic weight：Image body portion hsv color spatial model is extracted, using the method for homogenous segmentations
Feature is quantified, then by characteristic vector unification to a feature degree, is calculated between query image and image to be retrieved
The sequence of feature degree Euclidean distance and label, color characteristic weight is set up according to color notation；
S04, the selection for perceiving Hash system number：To improve the recall precision of large-scale image, the present invention is calculated using Hash is perceived
Method principle, vision word is quantified respectively to different Hash codes, correct with retrieval quantization time of the invention is considered
After rate, system number of the invention is chosen for 4；
S05, image similarity are calculated：The perceived distance of image Hash sequence is calculated by Hamming distance, then color characteristic is weighed
Recast obtains similarity sim (q, the d)=C of final query image and image to be retrieved for the weight coefficient of Hamming distanceqdHqd,
Will similar image output the most.
2. the vehicle image search method of Fusion of Color feature according to claim 1 and words tree, the step S02 bags
Include：
(1) to extracting body portion first per piece image in test library, SIFT feature is extracted to body portion, thus obtains
One characteristic set F={ f (i) } and corresponding image ID set imgID={ id (i) }；
(2) SIFT feature is distributed to by closest vision word in words tree, such each image according to closest principle
SIFT feature be converted to vision word；
(3)wi,jThen represent image djMiddle vision word FiWeights, i.e. vision word FiWith image djDegree of correlation, weights are
What the principle according to TF-IDF was defined, mi,jRepresent vision word FiIn image djThe number of times of middle appearance, it can be used to measurement and regards
Feel that word is used for describing the fine or not degree of image, N represents the sum of image in image library, niRepresent and include vision word FiFigure
As number, inverse document frequency is defined as idfi=lg (N/ni), it represent be meant that vision word for distinguish similar image and
The effect size of dissimilar image, so, wi,jCan be expressed as
(4) image djD can be expressed as with the weight vector of vision wordj=[w1,j,w2,j,L,wt,j], in test image storehouse
All images can be expressed in matrix as
3. the vehicle image search method of Fusion of Color feature according to claim 1 and words tree, the step S03 bags
Include：
(1) HSV space color feature vector is directly extracted to image, in order to reduce computational complexity, using homogenous segmentations
Method quantifies to color characteristic, in order to reduce computational complexity, color characteristic is carried out using the method for homogenous segmentations
Quantify, according to formula L=HQSQV+SQV+ V by characteristic vector unification to a feature degree, wherein QS、QVSaturation is represented respectively
The quantization series of degree and luminance component；
(2) the feature degree Euclidean distance between query image and image to be retrieved is calculated, is carried out according to Euclidean distance is ascending
Sort and label, color it is most like marked as 1, second is similar marked as 2, by that analogy；
(3) label according to color sets up color characteristic weight, the color characteristic weight calculation formula of query image q and image d
ForWherein RqdRepresent when query image is q, the label of image d.
4. the vehicle image search method of Fusion of Color feature according to claim 1 and words tree, the step S04 bags
Include：
(1) in view of system number of the invention is 4, the sum of same vision word weights in each image is calculated, averagely divides it
It is four sections, each segment table shows a Hash codes, obtains four Hash codes intervals；
(2) the Hash codes interval of the corresponding vision word of each vision word is compared, respectively by each vision list
Word is mapped to Hash codes, obtains the Hash codes sequence of image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611069889.4A CN106776856B (en) | 2016-11-29 | 2016-11-29 | Vehicle image retrieval method integrating color features and vocabulary tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611069889.4A CN106776856B (en) | 2016-11-29 | 2016-11-29 | Vehicle image retrieval method integrating color features and vocabulary tree |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106776856A true CN106776856A (en) | 2017-05-31 |
| CN106776856B CN106776856B (en) | 2020-07-03 |
Family
ID=58905345
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611069889.4A Active CN106776856B (en) | 2016-11-29 | 2016-11-29 | Vehicle image retrieval method integrating color features and vocabulary tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106776856B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107729379A (en) * | 2017-09-15 | 2018-02-23 | 中原智慧城市设计研究院有限公司 | Particular vehicle search method based on deep learning algorithm |
| CN108108657A (en) * | 2017-11-16 | 2018-06-01 | 浙江工业大学 | A kind of amendment local sensitivity Hash vehicle retrieval method based on multitask deep learning |
| CN109634953A (en) * | 2018-11-07 | 2019-04-16 | 宁波大学 | A kind of weight quantization Hash search method towards higher-dimension large data sets |
| CN110458175A (en) * | 2019-07-08 | 2019-11-15 | 中国地质大学（武汉） | It is a kind of based on words tree retrieval unmanned plane Image Matching to selection method and system |
| CN110490789A (en) * | 2019-07-15 | 2019-11-22 | 上海电力学院 | A kind of image hashing acquisition methods based on color and structure feature |
| CN110969202A (en) * | 2019-11-28 | 2020-04-07 | 上海观安信息技术股份有限公司 | Portrait collection environment verification method and system based on color component and perceptual hash algorithm |
| CN110968721A (en) * | 2019-11-28 | 2020-04-07 | 上海冠勇信息科技有限公司 | Method and system for searching infringement of mass images and computer readable storage medium thereof |
| CN112232428A (en) * | 2020-10-23 | 2021-01-15 | 上海电力大学 | Image hash acquisition method based on three-dimensional characteristics and energy change characteristics |
| CN112784632A (en) * | 2019-11-06 | 2021-05-11 | 济南信通达电气科技有限公司 | Method and device for detecting potential safety hazard of power transmission line |
| CN113626629A (en) * | 2021-08-17 | 2021-11-09 | 北京计算机技术及应用研究所 | Method for calculating image hash value by using scale invariant features |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020111A (en) * | 2012-10-29 | 2013-04-03 | 苏州大学 | Image retrieval method based on vocabulary tree level semantic model |
| US20140133759A1 (en) * | 2012-11-14 | 2014-05-15 | Nec Laboratories America, Inc. | Semantic-Aware Co-Indexing for Near-Duplicate Image Retrieval |
| CN104199922A (en) * | 2014-09-01 | 2014-12-10 | 中国科学院自动化研究所 | Large-scale image library retrieval method based on local similarity hash algorithm |
| CN106021251A (en) * | 2015-09-16 | 2016-10-12 | 展视网（北京）科技有限公司 | Hierarchical semantic model image retrieval method based on background knowledge |
-
2016
- 2016-11-29 CN CN201611069889.4A patent/CN106776856B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020111A (en) * | 2012-10-29 | 2013-04-03 | 苏州大学 | Image retrieval method based on vocabulary tree level semantic model |
| US20140133759A1 (en) * | 2012-11-14 | 2014-05-15 | Nec Laboratories America, Inc. | Semantic-Aware Co-Indexing for Near-Duplicate Image Retrieval |
| CN104199922A (en) * | 2014-09-01 | 2014-12-10 | 中国科学院自动化研究所 | Large-scale image library retrieval method based on local similarity hash algorithm |
| CN106021251A (en) * | 2015-09-16 | 2016-10-12 | 展视网（北京）科技有限公司 | Hierarchical semantic model image retrieval method based on background knowledge |
Non-Patent Citations (1)
| Title |
|---|
| 王娟等: ""基于颜色特征的图像检索技术"", 《计算机系统应用》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107729379A (en) * | 2017-09-15 | 2018-02-23 | 中原智慧城市设计研究院有限公司 | Particular vehicle search method based on deep learning algorithm |
| CN107729379B (en) * | 2017-09-15 | 2021-01-12 | 中原智慧城市设计研究院有限公司 | Specific vehicle retrieval method based on deep learning algorithm |
| CN108108657B (en) * | 2017-11-16 | 2020-10-30 | 浙江工业大学 | Method for correcting locality sensitive Hash vehicle retrieval based on multitask deep learning |
| CN108108657A (en) * | 2017-11-16 | 2018-06-01 | 浙江工业大学 | A kind of amendment local sensitivity Hash vehicle retrieval method based on multitask deep learning |
| CN109634953A (en) * | 2018-11-07 | 2019-04-16 | 宁波大学 | A kind of weight quantization Hash search method towards higher-dimension large data sets |
| CN109634953B (en) * | 2018-11-07 | 2021-08-17 | 宁波大学 | Weighted quantization Hash retrieval method for high-dimensional large data set |
| CN110458175A (en) * | 2019-07-08 | 2019-11-15 | 中国地质大学（武汉） | It is a kind of based on words tree retrieval unmanned plane Image Matching to selection method and system |
| CN110490789A (en) * | 2019-07-15 | 2019-11-22 | 上海电力学院 | A kind of image hashing acquisition methods based on color and structure feature |
| CN110490789B (en) * | 2019-07-15 | 2023-03-28 | 上海电力学院 | Image abstract obtaining method based on color and structural features |
| CN112784632A (en) * | 2019-11-06 | 2021-05-11 | 济南信通达电气科技有限公司 | Method and device for detecting potential safety hazard of power transmission line |
| CN112784632B (en) * | 2019-11-06 | 2023-05-30 | 济南信通达电气科技有限公司 | Method and device for detecting potential safety hazards of power transmission line |
| CN110968721A (en) * | 2019-11-28 | 2020-04-07 | 上海冠勇信息科技有限公司 | Method and system for searching infringement of mass images and computer readable storage medium thereof |
| CN110969202A (en) * | 2019-11-28 | 2020-04-07 | 上海观安信息技术股份有限公司 | Portrait collection environment verification method and system based on color component and perceptual hash algorithm |
| CN110969202B (en) * | 2019-11-28 | 2023-12-19 | 上海观安信息技术股份有限公司 | Portrait acquisition environment verification method and system based on color component and perceptual hash algorithm |
| CN112232428A (en) * | 2020-10-23 | 2021-01-15 | 上海电力大学 | Image hash acquisition method based on three-dimensional characteristics and energy change characteristics |
| CN113626629A (en) * | 2021-08-17 | 2021-11-09 | 北京计算机技术及应用研究所 | Method for calculating image hash value by using scale invariant features |
| CN113626629B (en) * | 2021-08-17 | 2023-09-29 | 北京计算机技术及应用研究所 | Method for calculating image hash value by utilizing scale invariant features |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106776856B (en) | 2020-07-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106776856A (en) | A kind of vehicle image search method of Fusion of Color feature and words tree | |
| CN105630901A (en) | Knowledge graph representation learning method | |
| WO2018023734A1 (en) | Significance testing method for 3d image | |
| CN107247971B (en) | Intelligent analysis method and system for ultrasonic thyroid nodule risk index | |
| CN107122809A (en) | Neural network characteristics learning method based on image own coding | |
| CN106203483B (en) | A kind of zero sample image classification method based on semantic related multi-modal mapping method | |
| CN109344821A (en) | Small target detecting method based on Fusion Features and deep learning | |
| CN112434169A (en) | Knowledge graph construction method and system and computer equipment | |
| CN108875076B (en) | Rapid trademark image retrieval method based on Attention mechanism and convolutional neural network | |
| CN106126585B (en) | The unmanned plane image search method combined based on quality grading with perceived hash characteristics | |
| TW201324378A (en) | Image Classification | |
| CN108062421A (en) | A kind of extensive picture multiscale semanteme search method | |
| CN104008375A (en) | Integrated human face recognition mehtod based on feature fusion | |
| CN105184298A (en) | Image classification method through fast and locality-constrained low-rank coding process | |
| CN106528662A (en) | Quick retrieval method and system of vehicle image on the basis of feature geometric constraint | |
| CN105183795B (en) | Remote Sensing Imagery Change Detection information retrieval method based on content | |
| CN104809469A (en) | Indoor scene image classification method facing service robot | |
| CN106886745B (en) | A kind of unmanned plane reconnaissance method generated based on real-time online map | |
| CN106844524A (en) | A kind of medical image search method converted based on deep learning and Radon | |
| CN109635140A (en) | A kind of image search method clustered based on deep learning and density peaks | |
| CN104142995A (en) | Social event recognition method based on visual attributes | |
| CN108427745A (en) | The image search method of visual dictionary and adaptive soft distribution based on optimization | |
| CN112949740A (en) | Small sample image classification method based on multilevel measurement | |
| CN106156798A (en) | Scene image classification method based on annular space pyramid and Multiple Kernel Learning | |
| CN104484347B (en) | A kind of stratification Visual Feature Retrieval Process method based on geography information |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240606 Address after: 518000, 3rd Floor, Building 23, Science Park West, No.1 Kezhi West Road, Science Park Community, Yuehai Street, Nanshan District, Shenzhen City, Guangdong Province, China 355 Patentee after: Shenzhen Jilong Technology Co.,Ltd. Country or region after: China Address before: 518000 1104, Building A, Zhiyun Industrial Park, No. 13, Huaxing Road, Henglang Community, Longhua District, Shenzhen, Guangdong Province Patentee before: Shenzhen Hongyue Information Technology Co.,Ltd. Country or region before: China Effective date of registration: 20240606 Address after: 518000 1104, Building A, Zhiyun Industrial Park, No. 13, Huaxing Road, Henglang Community, Longhua District, Shenzhen, Guangdong Province Patentee after: Shenzhen Hongyue Information Technology Co.,Ltd. Country or region after: China Address before: No. 1800 road 214122 Jiangsu Lihu Binhu District City of Wuxi Province Patentee before: Jiangnan University Country or region before: China |